Table A4: Summary of Study Methodological Characteristics that Impact Study Quality^*.

Study	N	Adequate Randomization Methods	Adequate Allocation Concealment	Blinding	Power	Loss to FUP	Intention-to-Treat
Cotton et al, 2000 (21)	81	✓ (random numbers)	✓	NR	No a priori sample size calculation Underpowered based on post hoc sample size calculations	✓	Methods say ITT but does not account for withdrawals
Davies et al, 2000 (20)	150	Methods not reported	Unclear (opaque envelopes not specified)	NR	A priori sample size calculation Underpowered based on post hoc sample size calculations	✓	Methods say ITT but does not account for withdrawals
Ojoo et al, 2000 (22)	60	Methods not reported	Unclear (opaque envelopes not specified)	NR	No a priori sample size calculation Underpowered based on post hoc sample size calculations	20%	NR
Aimonino Ricauda et al, 2008 (19)	104	✓ (random numbers)	✓	✓ (outcome assessment blinded)	A priori sample size calculation Adequate power for readmissions Underpowered for mortality and other outcomes based on post hoc sample size calculations	✓	Methods say ITT but does not account for withdrawals
Shepperd et al, 1998, (15;18)	32†	✓ (computer-generated random numbers)	✓	NR	A priori sample size calculation (HRQOL outcome) Underpowered based on post hoc sample size calculations	NR	Methods say ITT but not clear if analysis accounts for withdrawals
Skwarska et al, 2000 (6)	184	✓ (computer-generated random numbers)	NR	NR	No a priori sample size calculation Underpowered based on post hoc sample size calculations	30%	NR

^*Abbreviations: FUP, follow-up; ITT, intention-to-treat; N, sample size; NR, not reported.

^†The total sample size of the Shepperd et al (15;18) study is 538, but only 32 of those are COPD patients and therefore included in this analysis.